Temporal logic control of general Markov decision processes by approximate policy refinement.
Sofie HaesaertSadegh SoudjaniAlessandro AbatePublished in: ADHS (2018)
Keyphrases
- markov decision processes
- temporal logic
- optimal policy
- policy evaluation
- policy iteration
- model checking
- infinite horizon
- markov decision process
- factored mdps
- finite horizon
- state space
- finite state
- partially observable
- reinforcement learning
- dynamic programming
- state and action spaces
- average cost
- transition matrices
- average reward
- reward function
- action space
- decision processes
- modal logic
- decentralized control
- decision problems
- decision theoretic planning
- least squares
- action selection
- discounted reward
- planning under uncertainty
- markov decision problems
- reinforcement learning algorithms
- stationary policies
- linear temporal logic
- long run
- model free
- partially observable markov decision processes
- temporally extended
- expected reward